Experimental Demonstration of Learned Time-Domain Digital Back-Propagation

This work addresses the challenge of reducing computational complexity in optical communication systems for improved efficiency, though it is incremental as it builds on existing DBP methods.

The paper tackled the problem of compensating nonlinear signal distortions in optical fiber communications by experimentally demonstrating learned time-domain digital back-propagation (DBP) for the first time, achieving performance gains comparable to conventional frequency-domain DBP with lower complexity in a 64-GBd dual-polarization 64-QAM signal transmission over 1014 km.