Signal demodulation with Deep Learning Methods for visible light communication

Demodulation techniques are extremely important to ensure successful reception. Improving demodulation performance also improves the efficiency of a communication system as a whole. Traditional demodulators, with their high cost of implementation and tedious development, require dedicated hardware platforms. A single architecture for automatically demodulated end-to-end signals is suggested in this work. It is a demodulation based on machine learning (ML) for visible light communication (VLC), this suggested receiver which is based on learning can perform demodulation without any change in the receiver hardware by loading data on the network. The modulated signals are demodulated using a CNN-based demodulator, which turns them into images and then recognizes them. and there is still The RNN architecture which is powerful in terms of time series modeling. This type of receiver can intelligently and flexibly process several types of modulated digital signals. The outcomes reveal that the demodulation performances of data-driven demodulators are almost the same as those of conventional receivers. A combination of the strengths of the two architectures offers a more accurate demodulator, which is proposed in this work. a hybrid CNN and RNN architecture has shown its success in terms of its performance.