KNN, k-Means and Fuzzy c-Means for 16-QAM Demodulation in Coherent Optical Systems

In this paper, Machine Learning (ML) techniques such as k-Nearest Neighbors (KNN), k-Means and Fuzzy c-Means (FCM) are implemented in coherent optical system with DSP-based receiver. Nyquist single carrier optical transmission at 32 Gbaud is simulated in VPIDesignSuite Software in co-simulation with Matlab. Simulations results shows gains up to 1 dB and 2 dB using ML techniques at 50 km of optical link with 100 kHz and 25 kHz of laser linewidth, respectively. Besides, it is demonstrated that ML techniques can be effectively used as a nonsymmetrical demodulation (NSD) method.