1. Introduction

Quantum communication is proven to be only unconditionally secure method for information exchange, which could well be the first commercial application of quantum information science. Due to the low channel loss, free-space optical channels become far superior over fiber links to achieve ultra-long-distance quantum communication. With the help of satellite, it is possible to realize quantum communication networks and tests of quantum foundations [1–3] on a global scale. Significant experimental efforts have been devoted to investigate the feasibilities of satellite-based quantum communications. With the development of technologies, a series of researches have extended the communication distance of quantum key distribution (QKD) [4], entanglement distribution [5], and quantum teleportation [5, 6] between fixed locations to 100-km scale. Furthermore, full-scale verifications of ground-satellite QKD have been reported by utilizing a moving platform on a turntable and a floating platform on a hot-air balloon [7].

In the meantime, besides the experimental efforts in the ground, satellite based quantum communication projects have been proposed by several different countries [8–10] and the launching time has been scheduled. Before the launching of the satellite, however, a direct study of the whole process of transmitting and detecting single photons from the satellite to the ground with the help of existing satellite or aircraft at realistic envirments is necessary but remains experimental challenging owing to the high background noise level from all stars and the difficulties of tracking and synchronization.

Along this direction, an EU joint group reported an interesting attempt to verify the efficiency of atmospheric transmission [11] by sending laser pulses from the ground station to the satellite Ajisai equipped with cube-corner retroreflectors and further detecting the reflected laser signal with a modified classical laser ranging system (LRS). However, the simulated “transmitting” source on the satellite contains more than 1000 photons per pulse, which is far beyond single-photon level as the prerequisite of quantum communication. Further more, even with such a source and the time bin of 5 ns used in the experiment [11] for analyzing data, the obtained SNR is less than 1:1, which is not sufficient to create any secure key. Thus such a study failed to address the problem that whether it is possible to create satellite-ground quantum channel in the presence of all stray light. Here, we report for the first time an experimental simulation of a quasi-single-photon transmitter on the satellite with an average photon number of 0.85 per pulse and a full divergence angle of 38 μrad sending to the ground. In the ground station, by utilizing the coarse tracking techniques developed in [7], taking advantage of high-repetition-rate pulses (76 MHz) and improving the synchronization accuracy, we succeed to detect the desired single photons passing through the 400-km free-space channel with a SNR of 16:1. Such a SNR is good enough to generate quantum links for unconditionally secure QKD [7]. Hence, our results represent an important step towards satellite-to-ground QKD.

2. Scheme

As shown in Fig. 1, our experimental system consisted of a tracking and pointing telescope, a high repetition rate laser, a receiving terminal for collecting and detecting single-photons and a data-acquisition system. The telescope employed the binocular structure which could achieve the separation of emitting and receiving photons. The transmitting and receiving telescopes, with diameter of 20 cm and 60 cm respectively, are separated by 30 cm. The main advantage of this structure is that we can modulate the transmitting optical signal, and provide multiple-measurement basis on quantum signal conveniently. On the other hand, in order to realize the independence and synchronization between detecting quantum signal and LRS signal, we located dichroic mirrors (DMs) in both transmitting and receiving optical path.

 

Fig. 1 The scheme of the single-photon link from satellite to experimental setup installed at the Shanghai Observatory. CHAllenging Minisatellite Payload (CHAMP) was a German satellite launched July 15th, 2000 from Plesetsk, Russia and was used for atmospheric and ionospheric research, as well as other geoscientific applications. It was covered by 2 cm-diameter retroreflectors on each side. A train of pulses of 3 ps duration, 702 nm wavelength, 0.4 nJ of energy and 76 MHz repetition rate are coupled with the LRS pulses before being sent toward the transmitting telescope with aperture of 20 cm. The neutral density (ND) filter is used to control the weak pulse light, and the chopper is used to filter the backscattering noise. After the beam spreading, a fraction of the beam in the uplink path irradiates the satellite CHAMP. The corner cubes on the satellite retro-reflect back to the Earth a small portion of the photons in the laser pulse (downlink), which is the single-photon channel. Some of the photons in the downlink path are collected by the receiving telescope, a reflecting telescope with aperture of 60 cm, and detected by SPCMs, placed behind polarization measurement devices and spectral filters. The transmitting telescope and the receiving telescope are separated by a distance of 30 cm. All relevant events in the time domain are recorded by time measurement system TDC, and then referred to as coordinated universal time.

Download Full Size | PDF

At the transmitting side, a laser beam (wavelength of 702 nm, repetition rate of 76 MHz, single-pulse energy of 0.4 nJ, and pulse width of 3 ps) was coupled by a dichroic mirror with a green laser beam from the LRS of Shanghai Observatory. Then the coupled beams were transmitted from the transmitting telescope. Due to employment of a laser with high repetition rate, it made a possibility that we can receive enough single-photons coming from the satellites in a few minutes. The neutral density (ND) filter in the transmitting path was used to achieve energy control of the weak light pulse. The chopper was used to wipe out backscattering noise. Obviously, the telescope is composed by a number of mirrors, two of which control the position and height of the telescope. The rotation of the mirrors led to changes in polarization of the incident and the emergent light. An electronic-controlled half-wave plate (EC HWP) was added in the transmitter for polarization tracking.

At the receiving side, an optical detection was installed in the place of an original complementary metal-oxide semiconductor (CMOS). After the light reflected by the satellite enters the receiving telescope, it was focused by a primary and a secondary mirror. A meniscus lens reflected the light whose wavelength was 532 nm to the ocular. Because of the perfect separation of quantum signal and LRS signal, we could operate the complete polarization analysis of the incoming photons which will detected by single-photon counting modules (SPCMs). The transmitting and receiving photoelectric conversion signal would be led into a time-to-digital converter (TDC) and analyzed by a computer.

In this experiment, we have made some improvements to reduce background noise, such as adopting spatial, spectral and time filtering. We test the receiving field-of-view (FOV) of the detection system by scanning the telescope and recording the light intensity of the Polaris. As shown in Fig. 2, we experimentally demonstrated that FOV is 5″. Besides, a narrow band filter with the bandwidth of 3 nm was added to the receiving optical path to filter out background noise (including reflected lights from the sky and the ground) from the desired signal. In addition, we sectioned the transmitted pulsed laser from time sequence and filtered the backscattering by controlling the detection time. Every time after a transmitted laser pulse had passed the aerosphere, determining by the influence of the backscattering from the aerosphere was obviously within 30 km (round trip time of 200 μs), we detected the echo signal and avoided these kinds of noise. Besides, we did our experiment at night and choose the satellite in the shadow of the earth. Thus, even on a clear night we manage to reduce the background noise above the zenith angle of 45° to be less than 100 counts per second (cps).

 

Fig. 2 FOV of the detection system testing by scanning and recording the light intensity of the Polaris. The telescope was tracked to the Polaris, which could be seen as a fixed star. By changing the relative telescope tracking point, we record the different count rates of the changing position by SPCMs. Fig. 2(a) was the three-dimensional draw according to the counts. Fig. 2(b) was the fitting curve of FOV in the azimuth axial. Fig. 2(c) was the fitting curve of FOV in the zenith axial.The FOV of the receiving system, in both the azimuth and zenith axials, were measured to be less than 5″.

Download Full Size | PDF

The key experimental parameters were shown in table 1. Here, the pulse energy E was 0.4 nJ; S, the photon number per joule, was 3.53 × 10¹⁸ (702 nm); A_s, the effective area of satellite corner retro-reflector, was 11.34 cm²; A_r, the effective area of receiving telescope, was 0.25 m² ; K_t was the efficiency of transmission system of 0.2; K_r was the efficiency of optical receiving system of 0.15, including 0.5 for receiving telescope, 0.5 for coherent filter and 0.6 for multimode fiber collection; T was one-way atmospheric transmission of 0.6; η was the quantum efficiency of receiving detection device of 0.65; α was the attenuation factor of 13 dB, including 0.2 for the influence of satellite retro-reflector efficiency, 0.5 for both of atmospheric jitter and turbulence respectively; R was the satellite distance of 400 km for CHAMP; θ_t was the laser beam divergence angle of 300±15 μrad, as the LRS, where the 5 % uncertainty was mainly due to the temperature change of the environment (the telescope was designed for tracking satellites at different altitude from LEO to GEO, thus the divergence angle is far larger than the telescope’s diffraction limit at 400 km altitude); θ_s was the diffraction angle of satellite retro-reflector of 38 μrad, which was considered by both the clear aperture of the prism’s front face and the nominal width of the far field peaks given in [12, 13].

According to the parameters in Table 1, we could calculate the one-way channel attenuation from ground to satellite:

Table 1. Parameters of satellite-to-ground quantum channel link.

View Table

3. Result

In our experiment, the effective measurement time of the return signal of the CHAMP satellite (400 km) in each period (τ = 16 ms) of the chopped wave was τ₀ = 1.65 ms. For example, the number of echo signals in 10 s was 5385. Thus the average counts per second were $\overline{N}=5222$. The measured dark counts per second were $\overline{N_b}=89$. Therefore, the photon number per echo generated by a single laser pulse was experimentally measured:

The repetition rate of the present traditional LRS could be achieved at the magnitude of up to kHz. However, a light source having a higher repetition rate and time resolution was required due to the high loss in the channel. In the time synchronization system of our experiment, the quantum system and the LRS were independent. By this way we got rid of the frequency limitation of the LRS. We also utilized a TDC with high accuracy to record the arrival time of the photons and sample the pulsed photons [14]. After being reflected by the satellite and passing through a 400 km satellite-to-ground link, the photons would be detected. We defined t₀ as the emission time and t as the corresponding return time. According to the return time t of each photon, and the real-time satellite distance t_c, we could calculate the theoretical transmitting time t_exp, We used the difference D = t₀ − t_exp to characterize the accuracy of the time information.

Although the photons were emitted in pulsed intervals, the return time which we detected was disordered due to the changing distance of the satellite. To identify which pulse the photons belong to, we had to take the satellite distance measured by the laser as a synchronization data and obtained the round-trip time deviation by performing of offline analysis. The satellite orbit fitting curve was shown in Fig. 3. The distance of CHAMP satellite we measured in Shanghai Observatory was between 350km to 400km during our records. The histogram of all D values with 0.1 ns as the unit was shown in Fig. 4. The time synchronization accuracy was 1.35±0.03 ns at the full width at half maximum (FWHM). The main factors influencing time accuracy included: 350 ps for the time resolution of SPCMs, 160 ps for the accuracy of TDC, 1000 ps for the fitting precision of satellite orbit, and the rest effect for the time jitter of our pico-second pulsed laser.

 

Fig. 3 Plot of the range Rs between the satellite CHAMP and Shanghai Observatory. The round-trip time of the signal, directly proportional to R_s, exhibits strong and rapid variations during the satellite passage. The blue points are the LRS system’s records and the red line is the fitting curve on the survey data. The efficient time with maximal counts of single photons reflected from satellite and detected by SPCMs is shown in the gray box. The perigee height of CHAMP satellite that day was about 330km. The distance was ranging from 350 km to 400 km during our records of 15 s.

Download Full Size | PDF

 

Fig. 4 Histogram of all D values between expected and observed detections for CHAMP satellite.We summarized the D values numerically with 0.1ns as the unit. The peak of the histogram is centered at 0 ns, as expected. By Gaussian fitting, the full width at half maximum (FWHM) time accuracy was observed 1.35±0.03 ns.

Download Full Size | PDF

By using 2 ns bin size to deal with the experimental data, we got the effective echo signals counts $\overline{N^{\prime }}=1000$ and the effective dark counts $\overline{N_b^{\prime }}=58$. The SNR was calculated as follow:

The fitting precision was limited by the LRS’s frequency of 8Hz. If the frequency of the synchronization laser can be promoted up to KHz, the histogram would be much narrower (less than 1 ns). Under this modification, the signal-to-noise ratio will definitely be further improved.

In addition, the long-distance free-space quantum communication usually uses polarization coding. A crucial problem of satellite-to-ground communication was how to realize the polarization coding and basis measurement based on the kinematical reference system. While the satellite reference frame keeps changing during satellites’ motion, the satellite and ground are relatively static when the two sides stare at each other at any specific time in this frame. In our experiment, the reflecting surfaces of the retro-reflector were aluminium (Al) coated [12, 13]. Ground tests show that depolarization caused by Al coated reflector could not be compensated by unitary polarization transformation due to the different reflectivity on horizontal and vertical polarizations. Fortunately, this spacial problem will not exist in the final design [8]. As a demonstration, in the present experiment we achieve polarization maintaining by utilizing the reference frame.

As the telescope was composed by a number of mirrors, two of which control the position and height of the telescope. The rotation of the reflectors, which indicated by the azimuth and altitude encoder, led to changes in polarization [15, 16]. The real-time encoder data were used to drive the half-wave plate for polarization tracking. Fig. 5 shows the normalized intensity curve of horizontal and vertical polarization with the rotation of the telescope in the case of with or without polarization tracking. In case of no tracking, it looks like the sine curve. Finally, the maintenance of the polarization under tracking had been tested to be more than 100:1.

 

Fig. 5 Normalized intensity curve of horizontal and vertical polarization with the rotation of the telescope measured under the reference frame moving with the telescope. With a linearly polarized light incidenting without any compensation, the curve of horizontal (vertical) polarization show a sinesoid, as shown in curve H (V). With the polarization tracking system, the curve H′ and V′ show a nearly constant ratio. The maintenance of the polarization has a extinction ratio better than 100:1.

Download Full Size | PDF

4. Conclusion and perspectives

In summary, our experiment demonstrated the direct simulation of the single photon transmission through a satellite-to-ground free-space channel. From the quasi-single-photon transmitter on the CHAMP satellite, we observed the desired single photons with a counting rate up to 570 cps, a SNR of better than 16:1 and a time accuracy of 1.35 ns. These results are sufficient to set up an unconditionally secure QKD link between satellite and earth, technically [7]. In the scheduled Chinese Quantum Science Satellite [8], both the brightness of the light source and the divergence angle of the transmitter will be well improved and the fine acquiring, pointing and tracking techniques will be employed [5, 7]. Together with the field tests [5, 7], our results represent an crucial step towards the final implementation of high-speed QKD between the satellite and the ground stations, which will also serve as a test bed for secure intercontinental quantum communication.