MAXIMUM LIKELIHOOD CARRIER PHASE RECOVERY FOR LINEAR SUPPRESSED-CARRIER DIGITAL DATA MODULATIONS.

Abstract

The problem of maximum-likelihood ML estimation of the phase of a general data-modulated carrier is considered. The shortcomings of current iterative approaches to the problem are pointed out, and the correct conceptual approach is proposed. The true ML estimator is then obtained and found to be nonimplementable. However, by specializing to limits of high and low SNR, the general ML estimator is shown to reduce to implementable DA (decision-aided) and NDA (non-decision-aided) ML estimators, respectively. The DA receiver's performance in terms of phase tracking and symbol error probability can be analyzed, and even the effects of past decision errors on current system performance can be assessed. For circular signal constellations, the DA receiver has a simple and totally linear structure which is easy to implement. The NDA ML estimator is shown to be equivalent to the common carrier loops.