Machine learning based 64-QAM classification techniques for enhanced optical communication

Due to their greatly increased spectrum efficiency, high-order quadrature amplitude modulation (QAM) formats are especially successful at increasing transmission capacity. QAM is extremely sensitive to nonlinear distortion because of its dense constellation and SNR-hungry configuration. Autonomous neural network (ANN) derived nonlinear decision boundaries that are adaptively created by machine learning techniques can be used to classify symbols. The proposed work focusing on the quadrature amplitude modulation (QAM) scheme, the approach is to formulate an autonomous neural network (ANN) that can predict the class of each symbol from a signal stream of symbols. Experimental accuracy for each ANN's of proposed work achieves 89% by analysing all tests. Comprehensive results are presented with comparisons, demonstrating notable nonlinear mitigation with BER reductions. Additionally, it offers a glimpse into potential future research plans intended to raise the likelihood that predictions would come true and their accuracy.