VLC Color Shift Keying Detection Using Neural Networks

Abstract

In this work, a new Visible Light Communication system employing Color Shift Keying is presented, evaluated by means of a prototype. The variations of the ambient lighting constitute an interference at the detection of the symbols of the color constellation used in the transmitter. These variations are generally low speed compared to those due to data transmission. This paper provides two superimposed solutions for this problem: On the one hand, the RGB components of the detected image are compared with the temporary average extracted from the previous instants, which means that detection is not carried out using an absolute value. On the other hand, a neural network is used as a detector that through supervised training manages to distinguish the variations due to the transmission of those due to ambient light.