Fully-blind Neural Network Based Equalization for Severe Nonlinear Distortions in 112 Gbit/s Passive Optical Networks

This work addresses signal quality issues in high-speed optical networks, but appears incremental as it focuses on analyzing different neural network topologies without claiming major breakthroughs.

The paper tackled the problem of severe nonlinear distortions in 112 Gbit/s passive optical networks by developing a fully-blind digital signal processing chain, achieving a demonstration and evaluation for 100G PONs with low hardware complexity.