Deep Learning-Based Detector for Dual Mode OFDM With Index Modulation

Abstract

In this letter, we propose a deep learning-based dual mode orthogonal frequency division multiplexing with index modulation (DM-OFDM-IM) detector called DeepDM, which is close to optimal bit error rate (BER) performance with low computational complexity. DeepDM adopts a concatenation of convolutional neural network (CNN) and deep neural network (DNN) to detect index bits and carrier bits separately. A loss function is proposed to train the CNN and the DNN to approach the BER performance of the maximum likelihood detector. In addition, we propose a training method with selected data samples to make the neural networks converge fast. It is shown via simulations that DeepDM shows advantages over conventional detectors in terms of the BER performance and the computational complexity under the Rayleigh fading channel.