Application of frequency-domain features and high-order cumulants in ANN-based communication modulation recognition

The channel environment of wireless communication is growing more and more complex as communication technology advances, and modulation schemes are becoming more and more diverse. Traditional modulation recognition approaches will struggle to fulfill the future requirement for highly reliable and low latency wireless communication. Based on previous studies, this paper extracts the frequency domain features and high-order cumulants (HOC) of digitally modulated signals and analog modulated signals, and further builds the artificial neural networks (ANN) based communication modulation identification model. Then use it to classify the modulated signals to better cope with the diversity of communication systems. When the signal-to-noise ratio is 20 dB, the proposed method improves recognition accuracy by 18.82 percent compared to ANN-based signal modulation recognition without feature extraction.