Toward a Photonic Demonstration of Device-Independent Quantum Key Distribution

The security of quantum key distribution (QKD) usually relies on that the users’ devices are well characterized according to the security models made in the security proofs. In contrast, device-independent QKD—an entanglement-based protocol—permits the security even without any knowledge of the underlying quantum devices. Despite its beauty in theory, device-independent QKD is elusive to realize with current technologies. Especially in photonic implementations, the requirements for detection efficiency are far beyond the performance of any reported device-independent experiments. In this Letter, we report a proof-of-principle experiment of device-independent QKD based on a photonic setup in the asymptotic limit. On the theoretical side, we enhance the loss tolerance for real device imperfections by combining different approaches, namely, random postselection, noisy preprocessing, and developed numerical methods to estimate the key rate via the von Neumann entropy. On the experimental side, we develop a high-quality polarization-entangled photon source achieving a state-of-the-art (heralded) detection efficiency about 87.5%. Although our experiment does not include random basis switching, the achieved efficiency outperforms previous photonic experiments involving loophole-free Bell tests. Together, we show that the measured quantum correlations are strong enough to ensure a positive key rate under the fiber length up to 220 m. Our photonic platform can generate entangled photons at a high rate and in the telecom wavelength, which is desirable for high-speed generation over long distances. The results present an important step toward a full demonstration of photonic device-independent QKD.

Figure 1

An illustration of the device-independent QKD protocol. Alice and Bob share a pair of entangled photons potentially controlled by Eve (${\rho }_{ABE}$). Alice performs a measurement to her share with binary input $x\in \{1,2\}$ and binary output $a\in \{0,1\}$. Bob performs a measurement to his share with triple input $y\in \{1,2,3\}$ and binary output $b\in \{0,1\}$.

Figure 2

Schematic of the experiment. a: Entanglement source, creation of pairs of entangled photons: light pulses of 10 ns are injected at a repetition pulse rate of 2 MHz into a periodically poled potassium titanyl phosphate (PPKTP) crystal in a Sagnac loop to generate polarization-entangled photon pairs. The two photons of an entangled pair at 1560 nm travel in opposite directions to two sites Alice and Bob, where they are subject to polarization projection measurements. The PPKTP is placed in the middle of the hypotenuse of the Sagnac loop with a small angle to the light path, which does not significantly affect the upper limit of efficiency that the photonic setup could achieve, but can suppress the reflection of imperfect devices for 1560 nm photons. These enhancements lead to the nonmaximally entangled state generated in our experiment having a better fidelity $99.52\pm 0.15\%$ as compared to our previous work [26, 28, 29]. b: Alice and Bob, single-photon polarization measurement: in the measurement sites, Alice (Bob) uses a set of HWP and QWP to project the single photon into predetermined measurement bases. After being collected into the fiber, the single photons transmit through a certain length of fiber and then are detected by a superconducting nanowire single-photon detector (SNSPD) operating at 1 K. HWP, half-wave plate; QWP, quarter-wave plate; DM, dichroic mirror; PBS, polarizing beam splitter.