Application of optimization methods to minimize noise in quantum key distribution systems integrated into fiber-optic communication lines using dense wavelength division multiplexing

Abstract

Subject of study. This study examined the applicability of the simulated annealing optimization algorithm to the search for optimal configurations of quantum and information channels required for a successful quantum key distribution session considering grids with different frequency spacings between channels, such as 12.5, 25, 50, and 100 GHz. Aim of study. This study was aimed at optimization of a search for configurations of quantum and information channels for a grid corresponding to dense wavelength division multiplexing standards using optimization methods based on analysis of the channel noise effect on quantum key distribution sessions and investigation of these methods. Method. Naturally, quantum signal powers are significantly lower than those of classic signals. Therefore, noise from information channels for a configuration where they propagate in the same fiber with quantum channels results in considerable reduction of quantum key distribution system performance. This problem can be mitigated by reducing the noise level in the channel via the selection and use of an optimal spectral channel allocation. The search for such channel allocation schemes, which are further referred to as configurations, can be implemented via several methods, such as the educated guess method and the approach entailing optimization using the simulated annealing optimization algorithm, which are the subjects of this study. The results obtained using this method were compared with similar results obtained using the educated guess method. Descriptions of the methods and their interpretation for the context of the objectives of the study are presented together with mathematical models for calculation of noise resulting from spontaneous Raman scattering, four-wave mixing, and linear channel crosstalk. Optimal allocation schemes of quantum and information channels in the grid with dense wavelength division multiplexing (configurations) obtained using the two methods are described. The configurations were considered optimal when the total value of all channel noise types considered in the mathematical model was minimal. Main results. Optimal configurations obtained using methods of simulated annealing and educated guess are described in this study. We demonstrated that the educated guess method is applicable when the spontaneous Raman scattering is the primary contributor to the total losses. In this case, the information channels should be arranged according to the Raman scattering cross-section graph. However, when the contribution of four-wave mixing is comparable with or exceeds that of spontaneous Raman scattering, the simulated annealing algorithm is preferrable to search for the optimal configurations. In this case, information channels should be located at a distance from the quantum channel: the smaller the grid spacing, the further the information channels are positioned from the quantum channel to obtain optimal configurations. Increase in the number of channels results in reduction in this separation distance inside the grid with fixed channel spacing. Practical significance. Currently, practical implementations of fiber-optic communication lines are inextricably connected with multiplexing technologies, particularly, dense wavelength division multiplexing, to satisfy the constantly growing demand in increasing information capacity of communication channels and constructing quantum networks. Assignment of separate fibers for quantum key distribution systems is not optimal; thus, existing technologies of channel multiplexing must be integrated into the quantum key distribution systems. However, a rapid increase in the quantum bit error rate occurs owing to the presence of undesirable channel noise in fiber-optic communication lines, which eventually results in the impossibility of performing a successful quantum key distribution session. Therefore, accurate and profound analysis of the contributions of channel noise types is required to minimize this noise via the selection of an optimal configuration of classical and quantum channels on the frequency grid.