Statistical properties of moments of the phase component of a received signal have been used to design an algorithm for classifying the modulation type of M-ary PSK signals (Soliman and Hsue, 1992). In Soliman and Hsue (1992), the algorithm was developed by employing the Tikhonov function as an approximation of the phase distribution of received signals. In this paper we develop a new moment-based classification algorithm by using a Fourier series expansion of the exact phase distribution. This new algorithm achieves two improvements: it owns the better capability, especially at low CNR, for example, it offers improvement of 2 dB when the probability of misclassification is 0.01, and it offers simpler computation for the nth moment. © 1995.
Yang, Y., & Soliman, S. S. (1995). An improved moment-based algorithm for signal classification. Signal Processing, 43(3), 231–244. https://doi.org/10.1016/0165-1684(95)00002-U