Deep Learning-Based Anti-Narrowband Interference MSK Noncoherent Receiver

Narrowband interference (NBI), as kind of hostile frequency-domain interference, can seriously impair the bit error rate (BER) performance of minimum shift keying (MSK) noncoherent detection. To reduce the impact of narrowband interference on BER performance, MSK noncoherent receivers generally first perform interference suppression on received signals. However, existing MSK noncoherent detection algorithms do not consider the distortion of MSK signals caused by interference suppression, which limits the BER performance of the MSK communication system with noncoherent detection under narrowband interference. In order to solve this problem, a deep learning-based MSK noncoherent receiver (DL-MSKNCR) is proposed in this paper. This receiver contains an interference suppression subnetwork and an MSK noncoherent detection subnetwork. Through jointly training and optimizing, the MSK noncoherent detection subnetwork can effectively cope with the distortion of MSK signals caused by the interference suppression subnetwork. The simulation results show that DL-MSKNCD significantly improves the BER performance of the MSK communication system with noncoherent detection under narrowband interference.