Nonlinear equalization method based on Gaussian mixture model clustering algorithm for a coherent optical OFDM communication system

In this paper, an unsupervised clustering algorithm based on the Gaussian Mixture Model (UCGMM algorithm) for the coherent optical OFDM communication system is proposed to determine the constellation diagram. The purpose of nonlinear equalization of communication systems is achieved. In a back to back transmission system, compared to the Kmeans algorithm and the without any clustering algorithm, the UCGMM algorithm can obtain gains of approximately 0.6dB and 2dB respectively. For the cases of simulation in optical fiber transmission, the transmission distance of UCGMM algorithm is extended by 45km relative to the K-means algorithm, and 75km relative to without any clustering algorithm. In both cases, the effectiveness of the proposed UCGMM algorithm in nonlinear equilibrium is proved.