Nonlinearities Mitigation in Radio over Fiber Links for Beyond 5G C-RAN Applications using Support Vector Machine Approach

Machine learning (ML) methodologies gave an innovative and realistic direction to cope up with nonlinearity issues in fiber optics communication. In this paper, a 40-Gb/s 128-quadrature amplitude modulation (QAM) signal based Radio over Fiber (RoF) system is experimentally evaluated for 70 km of standard single mode fiber length which utilizes support vector machine (SVM) decision method to indicate an effective nonlinearity mitigation. The influence of different impairments in the system is evaluated that includes the influences of Mach-Zehnder Modulator nonlinearities, in-phase and quadrature phase skew of the modulator, input signal power and noise due to amplified spontaneous emission. By employing SVM, the results demonstrated in terms of bit error rate and eye linearity suggest that impairments are significantly reduced and licit input signal power span of 5dBs is enlarged to 15 dBs.