Maximum likelihood PSK classifier
P.C. Sapiano, John D. Martin · 2002
A method is presented for the classification of the number of levels on a PSK signal in additive white Gaussian noise (AGWN). The technique uses maximum likelihood principles on the baseband quadrature samples, and has the flexibility to incorporate an arbitrary number of PSK types. The classification performance is examined theoretically and is found to provide better performance than any of the other techniques known in the literature. This is compared in a graphical form with the qLLR and optimum phase methods. The improvement over the qLLR technique is seen to be marginal, but the sensitivity of the new technique due to parametric degradation is seen to be better in the cases examined. In order to improve the computational efficiency, simplifying approximations for the likelihood functions are implemented through the use of Pade approximations.