Published Paper Details:

This paper is designed to retrieve bits from OF DMT time-domains, without the use of explicit DFT/IDFT. Deep and complex DCCN network is developed. Both the frame sync and the PAPR (peak to average power ratio) reductions in a single pilot sequence are achieved in this study. The two systems considered are DCO-OFDM and O-FDM divided by the asymmetric based optical orthogonal frequency division (ACO-OFDM). Only strange indexed subcontractors can have the pilot symbol. The pilot signal from the receiver uses the pilot frame to detect the mirror symmetry of the pilot symbol from the synchronization algorithm. In order to optimize and approximate LMMSE and traditional CP enhancements, the DCCN receiver can exceed legacy channel estimates with varying delays and mobility on Rayleigh's fading channels. However, the proposed approach is currently not supported by popular deep learning platforms, which benefit from complex neural networks' expressive nature.