Quasi-hybrid likelihood modulation classification with nonlinear carrier frequency offsets estimation using antenna arrays

A quasi-hybrid likelihood ratio test (qHLRT) -based algorithm is proposed for linear modulation classification, with unknown carrier frequency offsets (CFO). A blind symbol-rate-sampling nonlinear least-squares (NLS) CFO estimator, which has the advantage of simplicity relative to other estimator that requires over-sampling, is incorporated in the qHLRT algorithm. The classifier is simple to implement yet provides accurate enough classification. A receive antenna array is added to further enhance the classification performance. Although the method can be applied to different linear digital modulation, we concentrate in this paper on M-ary QAM. Simulation results presented for classification of M-ary QAM signals under AWGN channel show that the qHLRT classifier with NLS estimator offers an efficient way to combat the sensitivity to unknown CFO of the average likelihood ratio test (ALRT) algorithm, and the performance can be further improved by spatial diversify