SIC-MIMO Systems with Neural Networks in OFDM-Based WDM VLCs

In this paper, we describe a wavelength-division multiplexing visible light communication system using two colored light-emitting diodes (LEDs) with similar emission wavelengths. A method using a neural network to cancel the intercolor interference caused by the spectral overlap of LEDs is proposed. The experimental results demonstrate that signal separation is possible with a bit error rate comparable to that of the pilot-based method.