A qHLRT Modulation Classifier with Antenna Array and Two-Stage CFO Estimation in Fading Channels

A likelihood ratio test (LRT) -based modulation classifier is sensitive to unknown parameters, such as carrier frequency offset (CFO), phase shift, etc. To better handle this problem, a robust antenna array-based quasi-hybrid likelihood ratio test (qHLRT) approach is proposed in this paper. A non-maximum likelihood (ML) estimator is employed to reduce the computational burden of multivariate maximization. A double CFO estimation scheme is also proposed, which increases the accuracy of CFO estimation. To deal with channel fading, maximal ratio combining approach is applied for CFO estimation as well as the computation of the likelihood functions. It is shown that when implementing with the nonlinear least-squares (NLS) phase parameters estimator and the method-of-moment (MoM) amplitude estimator, our scheme offers an effective way to recognize linear modulation formats with unknown parameters in fading channels.