Reservoir Computing Based on LED model for Signal Equalization in VLC-OFDM System

Visible Light Communication (VLC) is a promising technology for next-generation wireless communication, but signal distortion caused by device nonlinearity remains a challenge. However, traditional reservoir computing (RC) method is data-driven, which means its performance is directly influenced by the quality and quantity of the training dataset. In this paper, a model-driven RC method is proposed, which utilizes nonlinear LED model as the activation function for RC to improve the performance of VLC-OFDM system. The numerical simulation results show that the model-driven RC method significantly reduces the system bit error rate (BER) compared with the traditional RC equalization algorithm. Additionally, this algorithm can increase the system rate more than 25% while under the hard-decision forward error correction (HD-FEC) BER limited.