Machine learning of SVM classification utilizing complete binary tree structure for PAM-4/8 optical interconnection

Abstract

A machine learning method of effective nonlinear decision frame for PAM-N system based on support vector machine (SVM) using complete binary tree (CBT) structure is demonstrated in this work. The simulations results indicate improved performance by the proposed classifier which enhances the power sensitivity by 2-dB and 6-dB at the receiver side in 100-Gbps PAM-4 and PAM-8 systems respectively.