A spatiotemporal multi-channel learning framework for PCMA signal detection and recognition algorithm

Paired Carrier Multiple Access signal (PCMA) signals use the same frequency to transmit the modulation waveforms of both sides of the communication. With the characteristics of high-frequency band utilization and strong anti-interception ability, PCMA signals are widely used in military and important commercial communication systems. Blind separation of PCMA signals is an important part of radio spectrum monitoring, high-precision modulation parameter information is the necessary basis for separation. In order to process the PCMA signal normally, it is necessary to distinguish the PCMA signal from the common single-channel satellite digital modulation signal, and then identify the modulation mode of the PCMA signal. The detection and modulation identification of the PCMA signal provides the necessary guidance for parameter estimation and blind processing. In this paper, the detection and modulation identification of the PCMA signal with low SNR are implemented by using the multi-channel feature to extract the different feature information of the signal, adopting the idea of feature fusion, and using the time and space features of the signal simultaneously. Simulation results show that the proposed algorithm can be widely applied to the blind detection and modulation identification of the PCMA signal with BPSK, QPSK, 8PSK and their mixtures, and the classification performance is better than the traditional deep learning network based on CNN and the traditional detection modulation identification algorithm.