Extended 4-point approximation of the optimal QAM modulation detector

Maximum likelihood (ML) based modulation detector is optimal in the sense that the detection error probability is minimized if no prior probability of candidate modulations is available at the detector. However, the high computational complexity of the ML-based detector strongly limits its practical implementation. This contribution deals with a computational-efficient approximation of the ML-based method, which extends the existing 4-point approximation by utilizing the special arrangement of constellation points of square-formed quadrature amplitude modulation (QAM) schemes. Simulation results show that the extended 4-point approximation based detector is able to provide nearly optimal performance.