Comparison between two ICA methodologies to separate mixed signals obtained from a VLC channel

In this work the separation of signals is developed by the traditional ICA method and the ICA that incorporates a neural network. These mixed signals are obtained from a VLC channel, this signal after being received is processed by these two methods and it is determined which of the two ICA methodologies is the most efficient. The first method commonly called FastICA is based on the covariance of the Negentropy, while the second uses a neural network in conjunction with the covariance, this method is commonly called InfoMax. The results obtained for both methods are satisfactory, especially with the second one, confirming in this way also the reliability of the VLC (Visible Light Communication) channel.