Deep Learning for an Anti-Jamming CPM Receiver
Y. Qiu, Colin Brown, L. Li · 2019
A novel continuous phase modulation (CPM) receiver model is proposed in this paper that employs deep learning (DL) techniques to improve the signal recovery and synchronization performance under heavy jamming conditions for frequency hopping (FH) based waveforms. A hybrid deep neural network is used to implement the DL in the receiver model. The simulation results show that the proposed receiver with DL boosting is robust under tone jamming which is a worst case scenario for a CPM receiver. The model achieves 3 - 5dB improvement under single-tone jamming, in terms of bit-error-rate (BER), and 2dB improvement under multi-tone jamming, compared with a receiver without DL.