A Machine Learning Based Optimization Approach for Continuous‐Variable Quantum Key Distribution

Abstract In the continuous‐variable quantum key distribution (CV‐QKD) system, the secret key rate and transmission distance are important factors to evaluate the quality of the system. Numerical methods are commonly employed to calculate the secret key rate of QKD, but they often face challenges to find the optimal parameter to enhance the secret key rate. In this study, a machine learning based parameter optimization approach is proposed for the Gaussian‐modulated CV‐QKD system. The proposed approach adopts the genetic algorithm backpropagation (GA‐BP) based approach to enhance the secret key rate by optimize parameters of the CV‐QKD system. Furthermore, under certain assumptions, some theoretical results are established, which enables us to identify the maximum secret key rate under a given GA‐BP network. Based on these results, a bisection based method is proposed that can effectively find the optimal modulation variance and the corresponding maximum secret key rate. Finally, the simulation results show that the proposed approach can achieve a higher secret key rate than the conventional methods under the numerical settings.