Low-complexity Chromatic Dispersion and Polarization Mode Dispersion Joint Compensation for Coherent Optical Network Transmission

A lightweight CNN-BiLSTM hybrid neural network structure is proposed for jointly compensating chromatic dispersion (CD) and polarization mode dispersion (PMD) in coherent fiber optic communications. Simulation results show that the modeling algorithm achieves better BER performance than the conventional algorithm in low noise environments. Compared with the traditional compensation methods, the CNN-BiLSTM model provides a new approach for efficient signal compensation with fast response speed, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$44\%$</tex> lower computational complexity and memory resource saving.