ML estimator based on the EM algorithm for subcarrier SNR estimation in multicarrier transmissions

Abstract

In this paper, considering multicarrier transmissions, we present a maximum likelihood estimator of the subcarrier signal-to-noise ratio (SNR) based on the expectation-maximization (EM) algorithm. This new estimator is applicable to any linearly-modulated signal. It is a non-data-aided (NDA) method since no a priori knowledge is assumed about the transmitted data. The channel gains and phase distortions on the different subcarriers are assumed to be constant during the observation window, and the signal is assumed to be corrupted by additive white Gaussian noise (AWGN). The performances of our estimator are empirically assessed using Monte-Carlo simulations, showing that the new algorithm reaches the corresponding Cramér-Rao lower bounds (CRLBs) over a wide SNR range.