Identification of Shortwave Radio Communication Behavior Based on Autocorrelation Spectrogram Features

Cognitive communication behavior is becoming a research hotspot in the field of confrontation. In theory, behavioral intention noncooperating parties can be obtained by analyzing signals. Considering complexity actual electromagnetic environment, even when signal-to-noise ratio (SNR) low, certain accuracy still needs to guaranteed. this paper, according five types physical burst waveforms defined shortwave radio interoperability standard, signal feature extraction method based on autocorrelation spectrogram features proposed, and two-input convolutional neural network (CNN) for classification designed improve identification ability behavior. The experimental results illustrate that kinds behaviors accurately identified noise large. paper directly analyze through layer without demodulation, which has grasp station real time.