New key feature extraction algorithm based on third-order cumulant for modulation recognition

A set of values for identifying different types of modulations is developed. A classifier using this set of values can suppress the additive Gaussian noise and detect modulation signal features. Computer simulations of 6 types of band-limited digitally modulated signals corrupted by band-limited Gaussian noise sequences had been carried out to measure the performance of the developed algorithm. The proposed features were influenced slightly when the signal-to-noise-ratio(SNR) changed at 10 dR and 20 dB. The success rates of artificial neural network classifier with the third-order cumulant features were higher than that without these features at a SNR of 2 dB.