Robust Maximum Likelihood Classification of Quadrature Amplitude Modulation

This paper is concerned with amplitude-based maximum likelihood classification for quadrature amplitude modulation formats in the presence of frequency offset. We first propose a simplified classification algorithm that incurs a small loss in performance. A general performance analysis is then elaborated, which can achieve a higher computational efficiency, compared with the existing one. By integrating blind time synchronization and signal-to-noise (SNR) estimation, we further develop a robust classifier that is able to deal with some commonly unknown system parameters including frequency offset, phase shift, time delay, and SNR.