Real-World Two-Photon Interference and Proof-of-Principle Quantum Key Distribution Immune to Detector Attacks

Several vulnerabilities of single-photon detectors have recently been exploited to compromise the security of quantum-key-distribution (QKD) systems. In this Letter, we report the first proof-of-principle implementation of a new quantum-key-distribution protocol that is immune to any such attack. More precisely, we demonstrated this new approach to QKD in the laboratory over more than 80 km of spooled fiber, as well as across different locations within the city of Calgary. The robustness of our fiber-based implementation, together with the enhanced level of security offered by the protocol, confirms QKD as a realistic technology for safeguarding secrets in transmission. Furthermore, our demonstration establishes the feasibility of controlled two-photon interference in a real-world environment and thereby removes a remaining obstacle to realizing future applications of quantum communication, such as quantum repeaters and, more generally, quantum networks.

Figure 1

(a) Drift of differential arrival time. Variation of the arrival time difference of attenuated laser pulses emitted at Alice’s and Bob’s after propagation to Charlie. (b) Variation in the overlap of the polarization states of originally horizontally polarized light (emitted by Alice and Bob) after propagation to Charlie. Both panels include temperature data (crosses), showing a correlation between variations of indistinguishability and temperature. In addition, despite local frequency locks, the difference between the frequencies of Alice’s and Bob’s lasers varied by up to 20 MHz per hour (not shown).

Figure 2

Aerial view showing Alice (located at SAIT Polytechnic), Bob [located at the University of Calgary (U of C) Foothills campus] and Charlie (located at the U of C main campus). Also shown is the schematic of the experimental setup. Optically synchronized using a master clock (MC) at Charlie’s, Alice and Bob (not shown; setup identical to Alice’s) generated time-bin qubits at a 2 MHz rate encoded into Fourier-limited attenuated laser pulses using highly stable continuous-wave lasers at 1552.910 nm wavelength, temperature-stabilized intensity and phase modulators (IM and PM), and variable attenuators (ATT). The two temporal modes defining each time-bin qubit were of 500 ps (FWHM) duration and were separated by 1.4 ns. The qubits traveled through 12.4 and 6.2 km of deployed optical fibers to Charlie, where a $50/50$ beam splitter followed by two gated ($10\mu \mathrm{s}$ dead time) InGaAs single-photon detectors (SPD) allowed projecting the bipartite state onto the $|{\psi }^{-}⟩$ Bell state. (This projection occurred if the two detectors indicate detections with a $1.4\pm 0.4\mathrm{ns}$ time difference.) The MC, the polarization controller (POC), and Alice’s frequency shifter (FS) are used to maintain indistinguishability of the photons upon arrival at Charlie. These three feedback systems are detailed in the Supplemental Material [20]. The individual setups for measurements using spooled fiber (arrangement (i)) are identical.

Figure 3

(a) Measured error rates $e_{\mu \sigma }^z$ and $e_{\mu \sigma }^x$ for Alice and Bob, either both using signal intensity or both using decoy intensity as a function of total loss $l=l_A+l_B$ (in dB). We note that every other combination of intensities used in the decoy-state analysis requires Alice or Bob (or both) sending vacuum, and thus the error rate is 50% and not plotted. (b) Experimentally obtained and simulated secret key rates as a function of total loss $l=l_A+l_B$ (in dB), with $l_A\cong l_B$, for optimized mean photon numbers. Experimental secret key rates are directly calculated from measured gains and error rates using the decoy-state method [25] (see the Supplemental Material [20] for details). In both panels, the secondary $x$ axis shows distance, assuming a loss of $0.2\mathrm{dB}/\mathrm{km}$. Diamonds depict results obtained using deployed fibers [see Fig. 2a]; all other data were obtained using fiber on spools. Uncertainties (1 standard deviation) were calculated for all measured points, assuming Poissonian detection statistics. We stress that the simulated values, computed using our model [29], do not stem from fits but are based on parameters that have been established through independent measurements. Monte Carlo simulations using uncertainties in these measurements lead to predicted bands as opposed to lines (for more details, see the Supplemental Material [20]).