Low-complexity identifier for M-ary QAM signals

Automatic modulation classifier that reports the modulation type of a received signal blindly has found many applications in adaptive communications, software defined radio and non-cooperative communications. However, the majority of existing techniques does not include the classification of QAM signals that have been widely used in modem standards and high capacity radio systems. This paper developed a low-complexity Haar wavelet transform based QAM classifier by taking into account the ease of computation of the Haar wavelet transform. The number of multiple steps in the Haar wavelet transform magnitude was first detected through finding the numbers of peaks in the histogram of the Haar wavelet transform magnitude. It is then used for recognizing the size of the constellation. Simulations proved that the developed QAM classifier has high identification accuracy. It achieves more than 94% correct classification when SNR is 8 dB and is practically inerrable when SNR is greater than 12 dB.