Deep Convolutional Learning-Aided Detector for Generalized Frequency Division Multiplexing with Index Modulation
In this paper, a deep convolutional neural network-based symbol detection and demodulation is proposed for generalized frequency division multiplexing with index modulation (GFDM-IM) scheme in order to improve the error performance of the system. The proposed method first pre-processes the received signal by using a zero-forcing (ZF) detector and then uses a neural network consisting of a convolutional neural network (CNN) followed by a fully-connected neural network (FCNN). The FCNN part uses only two fully-connected layers, which can be adapted to yield a trade-off between complexity and bit error rate (BER) performance. This two-stage approach prevents the getting stuck of neural network in a saddle point and enables IM blocks processing independently. It has been demonstrated that the proposed deep convolutional neural network-based detection and demodulation scheme provides better BER performance compared to ZF detector with a reasonable complexity increase. We conclude that non-orthogonal waveforms combined with IM schemes with the help of deep learning is a promising physical layer (PHY) scheme for future wireless networks