Maximum likelihood SNR estimators for digital receivers

The problem of estimating signal-to-noise ratio (SNR) is investigated, both for static and slowly fading channels without memory. Maximum likelihood SNR estimators are derived for sampled signal processing receivers as well as continuous time signal processing receivers. It is shown that a maximum likelihood SNR estimator operating with digitally modulated signals is a decision-based structure that necessarily incorporates a digital data receiver. Based on this observation, structures of the estimators for two measures of SNR are presented.