Hybrid Neural Network Receiver for Direct Detection of Minimum-phase Signal

Inhalt:
We propose a hybrid neural network receiver for direct detection of minimum-phase signal, outperforming the Kramers-Kronig receiver. The results show that it reduces the required CSPR for HDFEC by 3 dB, lowers sampling rate for complex field reconstruction, and effectively compensates for propagation-related impairments.