Deep Neural Networks for Multicarrier Waveforms Classification in UAV Networks

During the last decade, the adoption of Unmanned aerial vehicles (UAV) for wireless networks has been rapidly growing for both civilian and military applications, due to their flexibility and mobility. In this paper, we consider their use as an aerial base station, then investigate the task of automatic signal recognition (ASR) through a deep learning (DL) approach, while considering spectrally efficient filtered multicarrier waveforms, including UFMC, F-OFDM, FBMC and GFDM compared to the legacy OFDM. This study has been done through series of simulations; in each one, different filter sizes, number of filters and number of convolutional layers have been applied. The best performance, with an overall accuracy of 0.93 at 2 dB signal to noise ratio (SNR), has been obtained when using smaller networks, with smaller filter sizes and higher number of filters.