Adaptive real time selection for quantum key distribution in lossy and turbulent free-space channels

The unconditional security in the creation of cryptographic keys obtained by quantum key distribution (QKD) protocols will induce a quantum leap in free-space communication privacy in the same way that we are beginning to realize secure optical fiber connections. However, free-space channels, in particular those with long links and the presence of atmospheric turbulence, are affected by losses, fluctuating transmissivity, and background light that impair the conditions for secure QKD. Here we introduce a method to contrast the atmospheric turbulence in QKD experiments. Our adaptive real time selection (ARTS) technique at the receiver is based on the selection of the intervals with higher channel transmissivity. We demonstrate, using data from the Canary Island 143-km free-space link, that conditions with unacceptable average quantum bit error rate which would prevent the generation of a secure key can be used once parsed according to the instantaneous scintillation using the ARTS technique.

Figure 1

Experimental setup: Alice is located at JKT observatory in La Palma. Qubits are prepared using a field-programmable gate array (FPGA) controller, linear optical components, and attenuated lasers. Alice telescope is also configured to acquire the beacon beam sent by Bob, located at the Optical Ground Station in Tenerife, and used for tracking the transmitter in order to stabilize the pointing.

Figure 2

Comparison between the counts detected by the SPAD [green (dark) line] and the voltage measured by the fast photodiode at the receiver [red (light) line]. In the inset we show a zoomed detail of the acquisition (between $4.48\mathrm{s}$ and $5.54\mathrm{s})$ in order to better appreciate the correlation between the quantum and classical signals. We chose a particular inset but in all the acquisitions the two signals are correlated.

Figure 3

Mean counts per packet $\overline{\mathcal{S}}\left(V_{\mathrm{T}}\right)$ (normalized to the mean counts obtained without thresholding) and fraction of total count $S\left(V_{\mathrm{T}}\right)/S(V_{\mathrm{T}}=0)$ in function of the probe threshold.

Figure 4

Experimental occurrences of probe intensities (measured by photodiode voltages) and log-normal fit.

Figure 5

Experimental QBER $\left(Q_{\exp }\right)$ and secure key rate $\left(R_{\exp }\right)$ in functions of the probe threshold (measured by the photodiode voltage). Dashed and solid lines show the corresponding theoretical predictions $(Q_{\mathrm{th}}$ and $R_{\mathrm{th}}$, respectively).

Figure 6

Comparison between the rates achievable by the ARTS, the postselection, and the standard QKD technique (no selection). We assumed that the channel QBER is $3\%$ and the log-normal parameter is $\sigma =1$, similar to the parameter we measured in the tested free-space channel. The parameter $m$ represents the mean received bits per coherence time of the channel.