Experimental Demonstration of Nonlinearity Compensation by Using SVM and Nonlinear Volterra Equalizer for 80 GBd DP-16-QAM Transmission

We propose support-vector machine (SVM) based detection and nonlinear Volterra equalization applied to a data center interconnect link. Among the various algorithms for machine learning an SVM can provide a more powerful way of learning nonlinear functions. With the use of suitable kernel functions, the SVM is able to reduce errors caused by nonlinear phase noise (NLPN), laser phase noise, IQ imbalances, etc. For an 80GBd DP-16QAM modulated signal the SVM shows robustness against IQ-imbalances up to 43°-phase mismatch.