A Flexible Multi-Rate Adaptive Blind Equalization Scheme Based on Constellation Probability Shaping in Coherent Optical Fiber Communication Systems

A flexible multi-rate adaptive blind equalization scheme based on probabilistic shaping (PS) in coherent optical fiber communication systems is proposed, addressing the performance degradation of clustering algorithms when performing nonlinear blind equalization for high-order modulation signals in PS systems. The scheme analyzes received signals distribution using the discrete Gaussian fitting theory and combines it with a weighted clustering algorithm to implement the Gaussian-weighted decision equalization (GWDE) method, enabling adaptive blind equalization for PS signals. Moreover, the scheme utilizes the probability distribution characteristics of the equalization process to implement the shaping degree estimation (SDE), enabling accurate quantification of the signal shaping degree. The proposed scheme has been experimentally and numerically validated in a 64QAM coherent optical fiber communication system. The results show that for $\nu = 0.05$ PS-64QAM signals, the proposed scheme improves tolerance to nonlinear effects caused by launched optical power (LOP). Compared to the unequalized signal, tolerance to these effects increases by over 1 dBm; relative to the K-means equalized signal, tolerance to LOP nonlinear effects improves by approximately 0.7 dBm. Additionally, the scheme reduces computational complexity by 12% compared to K-means. The proposed scheme also achieves 100% accuracy in estimating the degree of signal shaping, given the prerequisite of ensuring reliable communication. This approach holds significant potential for the dynamic rate transmission of diverse fine-grained services in future optical fiber communication systems.