QAM constellation classification based on statistical sampling for linear distortive channels

The classification method for phase-amplitude-modulated signals transmitted through linear, frequency-selective fading channels is presented. Utilizing a Monte Carlo (MC) sampling technique, QAM constellation classification is based on Gibbs sampling. Constellation classification is achieved along with joint estimation of unknown system parameters, including channel taps and data symbols. When all channel parameters are known, direct steps are proposed for classification and data estimation without altering the general approach. Moreover, a multistage classification algorithm is also presented to improve computational efficiency. The proposed QAM classification methods are shown to perform well on various constellations with different cardinalities, as well as constellations with symbols.