Spatial Superimposition Based PAPR Reduction for UACO-OFDM Systems with Multiple LEDs

We propose spatial superimposition (SS) structures to reduce peak-to-average power ratio (PAPR) for unipolar asymmetrically clipped optical-orthogonal frequency division multiplexing (UACO-OFDM) light fidelity (LiFi) systems with multiple light emitting diodes (LEDs) at the transmitter. The traditional μ-law companding method only increases the small amplitudes of signals, while maintaining the maximum value of the signal, which provides the limited PAPR reduction. Hence, we propose an improved nonlinear μ-law companding approach for PAPR reduction by enhancing small-amplitude signals and compressing large-amplitude signals. Linear compression is further used in the transmitted signals to reduce the impact of LED nonlinearity. Multiple LEDs are utilized to compensate for signal distortion caused by joint linear and nonlinear compressions, requiring no decompanding as in the traditional method. However, there are a few negative compensation signals that can be made to be positive by adding a small value of direct current (DC) bias theoretically derived in a closed form. Also, we propose an enhanced SS (eSS) structure, where the turn-on and maximum voltages of LED are jointly considered in the PAPR reduction, requiring no additional DC bias. This is the first work to investigate channel diversity in the proposed structures, while the multiple channels are assumed to be highly correlated with each other due to small LED separation in the previous works. We propose a frequency-domain channel filling (FCF) approach and a time-domain CF (TCF) approach, to mitigate the channel differences, which enables effective equalization of received signals. This is also the first work to investigate the analytical bit error rate (BER) of UACO-OFDM systems with clipping noise. We derive the analytical BERs of the proposed SS and eSS structures, respectively. Simulation results verify the proposed approaches.