Low Complexity Convolutional Neural Networks for Equalization in Optical Fiber Transmission

This addresses signal degradation in optical fiber communication, offering a more efficient solution for the telecommunications industry, though it appears incremental as it builds on existing equalizer methods.

The paper tackles fiber transmission impairments by proposing a convolutional neural network equalizer, achieving a five-fold reduction in trainable parameters and a 3.5 dB improvement in MSE compared to digital backpropagation with similar complexity.