Optical orthogonal frequency division multiplexing (OOFDM) is a promising technology in the next generation of high-speed and long-haul optical transmission, due to its high spectral efficiency, high speed of data transmission and strong ability of anti-dispersion. But optical OFDM system has a very high peak-to-average power ratio (PAPR). High PAPR will bring instantaneous high optical power to the optical OFDM system. Asymmetrically clipping and signal scrambling based on fast Hartley transform for PAPR reduction is proposed in optical OFDM system. Firstly, IFFT/FFT module in each sub-block of traditional signal scrambling technique is replaced with inverse fast Hartley transform (IFHT) and fast Hartley transform (FHT) module, which yield to the real signal in OOFDM system. Then, asymmetrically clipping technique is applied to turn it into a positive and real signal. Finally, the signal with the minimum PAPR is selected for transmission in the fiber channel. The PAPR of the optical OFDM signal can be reduced effectively. And without the Hermitian symmetry, the space and computational complexity are reduced accordingly.
The paper considers adaptive blind equalization problem of multichannel systems in digital communication. A feedforward neural network with lateral connections is introduced as the equalizer to estimate the source symbols from the received signals only. The lateral connections of the network are able to perform the self-orthogonalization, and thus the network can improved the performance. Simulation results have demonstrated promising performance of the proposed neural approach.