Abstract:
Automatic identification of the modulation type of a signal is a rapidly evolving area. It is
simply an intermediate step between the processes of signal detection and demodulation.
^ However, QAM classification becomes challenging when increasing the number of points
in the signal constellation. Therefore, an attempt is made to investigate performance of
quadrature amplitude modulation (QAM) signals with higher order cyclic cumulants (CCs)
under probability of correct classification (Ptt) measurement within a pattern recognition
framework. The theoretical and simulation results are shown by choosing the feature
vector, a combination of fourth and sixth order CCs. A comparison is also made with
Pcc values by choosing fourth order CC alone. The proposed feature shows much higher
performance especially at low signal-to-noise ratio. More importantly, the combination
feature vector selected is robust to phase noise influence and carrier frequency offset.