A Near-Optimal Deep Detector for BER Minimization in an OFDM AF Relay System

This paper investigates a Near-Optimal Deep Learning (DL)-based detector for an Orthogonal Frequency Division Multiplexing (OFDM) Amplify and Forward (AF) relay system. The main goal of the proposed DL detector is to achieve the best Bit Error Rate (BER) performance by a low-complexity Deep Neural Network (DNN). A pre-processing stage based on the frequency domain is applied before entering the input data into DNN. Since the performance of the model is very sensitive to tuning the parameters, simulation results compare the BER performance for different scenarios to obtain an accurate model. Finally, we show that the desired model is close to the Maximum Likelihood (ML) as an optimal detector.