Optical-computing-based fiber nonlinearity compensation using a 2×4 micro-ring array

Abstract

This paper presents an optical-computing-based approach for compensating nonlinear impairments in optical fiber communication systems. By leveraging a $2\times 4$ electro-optically modulated micro-ring array, the proposed method computes the product of three real numbers in a single operation, thereby enabling efficient triplet generation for perturbation-based nonlinear compensation. This approach can significantly boost computational efficiency and lower energy consumption. Our results show that optical-computing triplets provide Q-factor improvements on par with those achieved by electrical-computing triplets, with errors following a normal pattern. The scalability of the architecture is examined for higher-dimensional operations. Despite inherent errors, the artificial neural network can effectively compensate for these discrepancies, demonstrating the excellent robustness and practical value in high-capacity optical networks.