Continuous-variable quantum secure direct communication (CV-QSDC) utilizes quadrature components of the quantized electromagnetic mode to accomplish direct transmission of secret messages over the channel, offering the advantage of its compatibility with telecommunication facilities. However, phase noises during the running of systems destroy the correlation between the data of two parties and severely limit the effective communication rate. In this paper, we design a phase reference scheme to achieve phase synchronization between the signal light and the local oscillator in the coherent-state CV-QSDC system. To deal with the inaccurate modeling of phase drift in reference pulses, we further propose a phase compensation scheme using model-based deep learning, in which KalmanNet learns the Kalman gain from the data to track the true values of phase drift. In the simulation experiment, the frequency noise acts as the source of inaccuracy, and the results show that the phase compensation scheme based on KalmanNet accomplishes phase estimation effectively and thus improves the communication performance of CV-QSDC.

Close

PrevNext

View figure in article

• Deep learning
• Optical quantum information processing
• Quantum cryptography
• Quantum information processing with continuous variables
• Quantum protocols
• Quantum states of light