Classification error probability of cosine moment method with small signal rejections

In this paper, QAM classification based on cosine moments is analyzed. The cosine moments give poor performance for QAM classification because of a large phase noise of small amplitude signals. In the method, these small amplitude signals are not employed for the classification. An analytical approach for the classification error probability is presented for the cosine moment method. Although small signal rejections give a nonlinear effect at the receiver, the analysis includes the effect and an exact estimation of moments for the small signal rejections. The cosine moments are derived in a form of the infinite series of elementary functions to give classification error probability.