Machine learning aided carrier recovery in continuous-variable quantum key distribution

Abstract The secret key rate of a continuous-variable quantum distribution (CV-QKD) system is limited by excess noise. A issue typical to all modern CV-QKD systems implemented with reference or pilot signal and an independent local oscillator controlling the noise generated from frequency phase accrued transmitter receiver. Therefore accurate estimation compensation, so-called carrier recovery, critical subsystem CV-QKD. Here, we explore implementation machine learning framework based on Bayesian inference, namely unscented Kalman filter (UKF), for compare it standard method previously demonstrated method. Experimental results obtained over 20-km fibre-optic link indicate that UKF can ensure very low even at powers. measurements exhibited variance high stability in wide range ratios. This may enable hardware complexity which seamlessly work diverse transmission lines.