Binary Signaling for Optical Wireless Channel based on Deep Learning

We propose a kind of parallel binary signaling for optical wireless channel based on deep learning. We develop an autoencoder structure for an optical wireless channel and it produces learned patterns of 2×2 light-emitting diode array for the binary signaling in the optical wireless channel, which is modeled with additive white Gaussian noise. We report the influence of received noise on learned pattern, which is assumed to experience variation of channel characteristics.