Amplitude-based neuro-classifier for classification of digital quadrature and staggered modulations

Recognition of the modulation type has some commercial and military applications. In this paper, we present a new classifier to recognize QPSK, SQPSK and MSK modulations from each other. This classifier, its main part is a neural network, makes use of two features that are based on the amplitude characteristics of the considered modulations. The employed features are easily computed. In the proposed classifier, no synchronization with arrival time of the received signal is needed. If the SNR of the transmitted signal is kept fixed in transmitter, the proposed classifier will be insensitive to the exact value of channel noise power. These properties make our classifier very robust