Robust continuous-variable quantum key distribution against practical attacks

Recently, several practical attacks on continuous-variable quantum key distribution (CVQKD) were proposed based on faking the estimated value of channel excess noise to hide the intercept-and-resend eavesdropping strategy, including the local oscillator (LO) fluctuation, calibration, wavelength, and saturation attacks. However, the known countermeasures against all these practical attacks will inevitably increase the complexity of the implementation of CVQKD and affect its performance. We develop here an asynchronous countermeasure strategy without structural modifications of the conventional CVQKD scheme. In particular, two robust countermeasures are proposed by adding peak-valley seeking and Gaussian postselection steps in conventional data postprocessing procedure. The analysis shows that the peak-valley seeking method naturally make the schemes immune to all known types of calibration attacks even when Eve simultaneously performs wavelength or LO fluctuation attacks and exhibit simpler implementation and better performance than the known countermeasures. Meanwhile, since the Gaussian postselection is able to resist the saturation attacks, the proposed schemes are secure against all known types of practical attacks.

Figure 1

The description of calibration attack by manipulating the classical synchronous pulse. The upper plot shows the profiles of the trigger signals generated at Bob's side depending on the shape of the classical synchronous pulse, the solid and dotted curves denote the original signal and the attenuated one by Eve. The lower plot shows the differential signal from homodyne detection, where the variance of the output value of the homodyne detection depending on the time of measurement. The trigger is delayed $t_0$ such that the value of the signal at the output of homodyne detection is maximized.

Figure 2

Countermeasure against practical attacks with power meter based on peak-valley seeking and Gaussian postselection. PC is polarization controller, BS is beam splitter, PBS is polarization beam splitter, FM is faraday mirror, RNG is the random number generator, PM is phase modulator. The power meter is used to monitor the LO intensity in real time.

Figure 3

Time-domain shapes of the output pulse from the homodyne detection for the conventional synchronous (upper) and peak-valley seeking (lower) schemes. The arrows denote the sampling positions in time flow.

Figure 4

Secret key rates under collective attacks with block size of $1\times {10}^9$ and security parameter $1\times {10}^{-10}$. The upper curve corresponds to the secret key rate of the proposed countermeasure with power meter based on peak-valley seeking and Gaussian postselection. The lower curve corresponds the countermeasure by inserting an optical switch on Bob's signal path to perform real-time shot-noise measurement.

Figure 5

Countermeasure against practical attacks with dual-sampling measurement based on peak-valley seeking and Gaussian postselection. PC is polarization controller, BS is beam splitter, PBS is polarization beam splitter, FM is faraday mirror, RNG is the random number generator, PM is phase modulator. ADC2 is used to monitor the corresponding intensity of LO signal in homodyne detection.

Figure 6

Time-domain shapes of the output pulse from the homodyne detection and PIN detector for the dual-sampling measurement. The arrows denote the sampling positions in time flow.

Figure 7

The secret key rates under collective attacks with (the dashed curves, the three curves are almost overlapped) and without (the solid curves) dual-sampling measurement, when considering the finite sampling bandwidth effects. Curves from top to bottom represent $f_{\text{rep}}=10$, 5, and 2 MHz.