A Novel PAPR Reduction Scheme Using ISLM Cascading Neural Network-Based Companding Algorithm for CO-OFDM Systems

In coherent optical orthogonal frequency division multiplexing (CO-OFDM) systems, the high peak-average-power ratio (PAPR) is the major disadvantage that impacts the system performance, which is manifested in limiting the maximum power of lasers and exacerbating the fiber nonlinear effects. Therefore, a novel PAPR reduction scheme using improved selective mapping (ISLM) cascading neural network (NN)-based companding algorithm is proposed. Compared with the original OFDM and the single traditional scheme, the proposed scheme shows better PAPR reduction effect and bit error rate (BER) performance. Compared with the cascaded scheme, the proposed scheme has the similar effect but lower complexity.