Robust QAM classification by moments and its error probability analysis

In this paper, a robust QAM classification in a frequency offset channel is considered. When a carrier frequency is not recovered with the sufficient accuracy in a receiver, a signal point rotates in a vector diagram. This paper presents the classification method by the amplitude and squared cosine moments. In the method, phase shifts are employed for the squared cosine moments in order to change a square QAM constellation to a virtual circular constellation, which improves the classification performance. Simulation results of the classification error probability show the effectiveness of the proposed method for QAM classification in the presence of frequency offset. The classification error probability is also expressed by both amplitude and squared cosine moments which are derived from the probability density function of the received signal.