High-rate discretely-modulated continuous-variable quantum key distribution using quantum machine learning

Continuous-variable quantum key distribution [1] is designed to implement point-to-point secret key distribution, its security is guaranteed by the fundamental laws of quantum physics [2]. In a basic version of CVQKD [3], the sender, called Alice, encodes secret key bits in the phase space of coherent states and sends them to an insecure quantum channel, while the receiver, called Bob, measures these incoming signal states with coherent detection. After several steps of data post-processing, a string of secret keys can be finally shared by Alice and Bob. One of the advantages of CVQKD is that it is compatible with most existing commercial telecommunication technologies [4-6], making it easier to integrate into real-world communication links. In general, a CVQKD system is mainly composed of quantum signal processing and data-postprocessing [7]. The former part corresponds to signal modulation, transmission, and measurement, aiming to generate a raw key, while the latter part corresponds to data reconciliation, parameter estimation, and privacy amplification, attempting to extract the final secret key from the raw key. In CVQKD, secret key rate and maximal transmission distance are generally a pair of crucial performance indicators. For a specific CVQKD system, however, there is tradeoff between the secret key rate and the maximal transmission distance: the longer the transmission distance, the lower the secret key rate, and vice versa. The main reason is that the continuous-variable quantum signal used to carry the secret key is extremely weak.