The CESAE multiple objection optimization network of the ACO-OFDM VLC system

Visible light communication (VLC) is a foreground technology in the sixth generation (6G) communication system. Simultaneously deep learning network is a novel approach to improve the properties of VLC system. In this paper, we put forward an asymmetric clipping optical orthogonal frequency division multiplexing (ACOOFDM) system based on the constant envelope stacked autoencoder (CESAE) network, which is a combination method of phase modulated constant envelope OFDM (CE-OFDM) and the SAE deep learning network. The CESAE network systematically improves the character of the whole system through multiple layers and the loss function. It is shown by the simulation results that tuning the phase modulation index can availably maintain the peak -to -average -power ratio (PAPR) to about 3 dB. In addition, a detailed comparison with traditional method reveals that the presented method fulfils superior BER performance and spectral efficiency.