Nonlinearity Mitigation for VLC with an Artificial Neural Network Based Equalizer

Light emitting diode (LED) is a major source of nonlinearity in visible light communication (VLC). In this work, we introduce a Wiener-Hammerstein model considering the LED nonlinearity and memory effect of an indoor channel. The VLC nonlinearity belongs to dynamic-range-limited nonlinearities. Considering the impact of nonlinearity strength in different dynamic regions, we design three types of signals for both weak and strong nonlinearity regions. Moreover, in order to well compensate the nonlinearity, an artificial neural network (ANN) based equalizer is compared with the conventional Volterra series-based equalizer and the memory orthogonal polynomial-based equalizer. The results show that the proposed equalizer significantly outperforms conventional nonlinear equalizers, up to two orders of magnitude in high SNR region.