Modulation Recognition of Satellite Communication Signals with Time-frequency Overlap Based on Improved High-order Cumulants

High-order cumulants (HOC), as a characteristic quantity that is insensitive to additive white Gaussian noise and can effectively represent the high-order statistical characteristics of signals, have become a key theoretical tool in the field of modulation recognition. However, most of the existing classical theoretical values of HOC are derived based on rectangular pulse shaping(RPS) signals, without considering the influence of other shaping methods, such as the raised cosine pulse shaping(RCPS) commonly used in satellite communication systems. The differences in the HOC characteristics between these two shaping methods have not been fully studied. Aiming at this problem, this paper proposes an improved HOC based on the time-frequency overlapping signal with raised cosine pulse shaping, constructs feature parameters according to its HOC characteristics, and finally combines the parameters to establish a decision tree for the recognition and classification of the mixed signal set. The simulation results show that the recognition rate of this algorithm for mixed signals is above $\mathbf{9 0 \%}$ when the signal-to-noise ratio(SNR) is greater than $\mathbf{5 d B}$.