Kernel-Based Learning-Aided Phase Noise Compensation in Dual-Pump Optical Phase Conjugation Coherent System

We deploy kernel-based time-series prediction to suppress the phase noise induced by small deviations from ideal pump counter-phasing in a dual-pump optical phase conjugation system. We show experimentally 1.5-dB SNR improvement for 16-QAM signals at 4° pump-phase mismatch.