A Model-Driven Deep Learning Method for LED Nonlinearity Mitigation in OFDM-Based Optical Communications

Abstract

The nonlinearity of light emitting diodes (LED) has restricted the bit error rate (BER) performance of visible light communications (VLC). In this paper, we propose model-driven deep learning (DL) approach using an autoencoder (AE) network to mitigate the LED nonlinearity for orthogonal frequency division multiplexing (OFDM)-based VLC systems. Different from the conventional fully data-driven AE, the communication domain knowledge is well incorporated in the proposed scheme for the design of network architecture and training cost function. First, a deep neural network (DNN) combined with discrete Fourier transform spreading (DFT-S) is adopted at the transmitter to map the binary data into complex I-Q symbols for each OFDM subcarrier. Then, at the receiver, we divide the symbol demapping module into two subnets in terms of nonlinearity compensation and signal detection, where each subnet is comprised of a DNN. Finally, both the autocorrelation of the learned mapping symbols and the mean square error of demapping symbols are taken into account simultaneously by the cost function for network training. With this approach, the LED nonlinearity and the interference introduced by the multipath channel can be effectively mitigated. The simulation results show that the proposed scheme exhibits better BER performance than some existing methods and further accelerates the training speed, which demonstrates the prospective and validity of DL in the VLC system.