Nonlinear compensation of OFDM Systembased on DNN

Aiming at the problem of nonlinear distortion caused by amplifiers during RF transmission in the additive white Gaussian noise channel, a nonlinear compensation method of orthogonal frequency division multiplexing (OFDM) based on deep neural network (DNN) is proposed. A two input and two output deep neural network with two hidden layers is designed to compensate the nonlinear distortion at the receiver, and the performance differences caused by different input and output data of the network are discussed. The simulation results show that this method can avoid the difficulty of signal acquisition and power loss caused by input back off, and can compensate nonlinear distortion to a certain extent.