Modeling of Data Transmission in Polymer Optical Fibers Using DMT and Machine Learning Techniques

Abstract

In local networks and other short links, polymeric optical fiber may be used due to its mechanical properties; however, some undesired and strong effects appear in these environments. This work proposes three different models to fine-tune the electrical parameters of discrete multi-tone (DMT) modulation to predict the signal-to-noise rate (SNR) over each subcarrier used with a 560 nm central wavelength. Considering the present setup, the devised models achieve some bold results, demonstrating the possibility of explaining more than 99% of the experimental data, in the best scenario. Several models are mathematically fit to obtain distinct adjustments for the nonlinearity present in the link considering the electrical parameters.