Decision Trees-based Demodulation in Gridless Nyquist-WDM Systems to Minimize ICI Effects

We propose a digital demodulation technique based on decision trees and random forest algorithms to minimize interchannel interference effects in gridless scenarios. We experimentally verified this technique in a 3×16-Gbaud 16-QAM Nyquist-WDM system. Gains up to 1.6 dB are achieved compared to conventional demodulation.