Señales DCO-OFDM eficientes para VLC utilizando Redes Neuronales adaptativas Difusas (ANFIS)

Abstract

In this paper, Adaptive Neural Fuzzy Inference Systems (ANFIS) are used for obtaining Direct Current biased Optical Orthogonal Frequency Division Multiplexing (DCOOFDM) signals with low power envelope. Signals with very low crest factor obtained from Active Constellation Expansion (ACE) method are used for training our fuzzy systems. After the off-line training, the ANFIS are able to generate from the original realvalued signal another real-valued one with more than 2 dB of power envelope reduction which impacts in the final performance after going through the LED.