Maximum-likelihood classification for digital amplitude-phase modulations

We apply the maximum-likelihood (ML) method to the classification of digital quadrature modulations. We show that under an ideal situation, the I-Q domain data are sufficient statistics for modulation classification and obtain a generic formula for the error probability of a ML classifier. Our study of asymptotic performance shows that the ML classifier is capable of classifying any finite set of distinctive constellations with zero error rate when the number of available data symbols goes to infinity.