Performance Study on Nonlinearity Distortion Mitigation in Modulated Optical Interconnects based on Machine Learning

In the recent days Machine Learning algorithms are widely used in the Optical domain. In this paper we have applied machine learning algorithm to mitigate the modulation nonlinearity distortions occurring in the Optical Interconnections. Optical interconnections are widely used due to the demerits of its electrical counterpart in terms of latency and power. A machine learning based detection scheme using complete binary tree Support Vector Machines (CBT-SVM) is proposed for the modulation nonlinearity mitigation and bit error rate (BER) estimation. An Optical interconnection link is modelled using the simulation setup in order to generate the datasets required for the experiment. A PRBS generator is used to modulate a VCSEL (Vertical Cavity Surface Emitting Laser) in order to produce PAM-4 signal. Controlled amounts of modulation non linarites can be introduced by varying bias currents and temperature of VCSEL. Various datasets were generated by varying these parameters. ML based detection scheme was employed using CBT SVMs and the bit error rates were estimated. The proposed technique has the potential to be used at the receiver side for intelligent signal analysis and optical performance monitoring. Also, we observed that by using CBT SVM we are able to achieve better BER (1e
^-8
) at improved data rates (10Gbps). The proposed model using CBT SVM machine learning algorithm can mitigate the modulation nonlinearity distortion.