PAPR Reduction Scheme for Localized SC-FDMA Based on Deep Learning

AbstractLarge peak-to-average power ratio (PAPR) hinders the development of the localized single carrier frequency division multiple access (SC-LFDMA). In this paper, autoencoder (AE) is introduced in SC-LFDMA to reduce PAPR, known as AE-SC-LFDMA. In AE-SC-LFDMA, the Encoder and Decoder of AE are used to encode and decode the modulated symbols of conventional SC-LFDMA based on deep neural network (DNN). This process aims to make AE-SC-LFDMA achieve lower PAPR as well as be more robust to the nonlinear distortion (NLD) of high power amplifier (HPA). Simulation results show that the proposed scheme outperforms conventional schemes both in bit error rate (BER) and PAPR.KeywordsSC-LFDMAAEDNNHPA