Noncoherent Maximum Likelihood Classification of Quadrature Amplitude Modulation Constellations: Simplification, Analysis, and Extension

Abstract

This paper is concerned with classification of quadrature amplitude modulation (QAM) signals by using noncoherent maximum likelihood (ML) schemes, which are robust to frequency mismatch and phase shift. To reduce the high complexity involved in optimal noncoherent ML classification, we propose two suboptimal alternatives that perform well under various conditions. In particular, one suboptimal scheme, derived from high signal-to-noise ratio (SNR) approximation, is shown to provide close-to-optimal performance. We also present a general performance analysis for noncoherent ML classification of K types of QAM constellations, ending up with simplified integration for K ; 5. In addition, we investigate the asymptotic behavior of noncoherent ML classification assuming the number of samples used increases to infinity. The asymptotic performance of optimal as well as suboptimal noncoherent ML classification is established.