Signal detection in OFDM systems for visible light communication based on deep learning

In this thesis, for signal detection in orthogonal frequency division multiplexing systems, the deep learning neural network is utilize to classify symbols at the receiver. The symbol error rate and least square and minimum mean square error estimations are compare after the long short-term memory based neural network has been train for a single subcarrier. This initial trial's offline training and online deployment stages are anticipate using a fixed wireless channel. Each transmitted orthogonal frequency division multiplexing packet has a random phase shift performed to assess the neural network's stability