Array-based Linear Modulation Classifier with Two-Stage CFO Estimation in Fading Channels

摘要

Likelihood ratio test (LRT) - based linear modulation classifier is sensitive to unknown parameters, such as carrier frequency offset (CFO), phase shift, etc. An antenna array-based quasi-hybrid likelihood ratio test (qHLRT) approach is proposed to cope with the problem. A non-maximum likelihood (ML) estimator is employed to reduce the computational burden of multivariate maximization. A two-stage CFO estimation scheme is also proposed to increase the accuracy of CFO estimation. To combat channel fading, maximal ratio combining (MRC) technique is applied for CFO estimation as well as the computation of the likelihood functions. The Cramer-Rao lower bound (CRLB) of the proposed CFO estimation method is derived. It is shown that with nonlinear least-squares (NLS) algorithm and method-of-moment (MoM) algorithm to estimate phase and amplitude respectively, our scheme offers an effective and practical solution to recognize linear modulation formats in fading channels.