Machine-Learning-Aided Optical OFDM for Intensity Modulated Direct Detection

End-to-end learning systems are conceived for Orthogonal Frequency Division Multiplexing (OFDM)-aided optical Intensity Modulation paired with Direct Detection (IM/DD) communications relying on the Autoencoder (AE) architecture in deep learning. We first propose an AE-aided Layered ACO-OFDM (LACO-OFDM) scheme, termed as LACONet, exploiting increased bandwidth efficiency of LACO-OFDM. LACONet employs a Neural Network (NN) at transmitter bit-to-symbol mapping, and another NN receiver recovering data bits, which together form AE can be trained end-to-end manner simultaneously minimizing both BER PAPR. Moreover, detection is drastically simplified compared to classical LACO-OFDM, since Fast Fourier Transform (FFT) operation applied only once receiver. further generalized OFDM scheme IM/DD communications, IMDD-OFDMNet, where unipolarity Time Domain (TD) signal no longer guaranteed by Hermitian Symmetry, but rather taking absolute square value complex TD signal. As such, all available subcarriers IMDD-OFDMNet used carrying useful information, hence it has higher throughput than LACO-based schemes. benefit, its transceiver requires single Inverse FFT or FFT. Finally, simulation results provided show that our schemes achieve better PAPR performance their conventional counterparts.