This paper considers the detection of coded phaseshift keying signals subjected to additive white Gaussian noise and oscillator phase noise. We propose a detector that partitions the received frame into smaller blocks and models the unknown phasor variations over each block as a truncated discrete cosine transform (DCT) expansion. Detection and decoding are iteratively performed between a soft-input soft-output (SISO) demodulator, a SISO demapper, and a SISO decoder based on the sum-product algorithm and the factor graph framework, whereas the expectation-maximization algorithm is used in the demodulator for the DCT coefficients estimation. The resulting demodulator is shown to have an excellent performance/complexity tradeoff and to be well-suited for parallel processing on multiple cores.

The presence of a frequency offset (FO) and phase noise can cause severe performance degradation in digital communication systems. This work combines a simple FO estimation technique with a low-complexity phase noise estimation method, inspired by the space-alternating generalized expectation-maximization algorithm. Using a truncated discrete-cosine transform (DCT) expansion, the phase noise estimate is derived from the estimated DCT coefficients of the phase. A number of implementations of the proposed algorithm are discussed. Numerical results indicate that when estimating the FO from pilot symbols only, comparable performance can be reached as the computationally more complex case where the FO is updated iteratively, with small convergence time. The phase noise estimation step is well capable of compensating for the residual FO. For the considered scenario, performing FO compensation before iterative phase noise estimation yields a bit-error rate performance degradation close to the case where the FO is known.

This contribution deals with estimation and compensation of phase noise in single-carrier digital communications. We present an iterative feedforward soft-decision-directed phase noise estimation algorithm, that is based on approximating the phase noise process by an expansion of DCT basis functions containing only a few terms. An initial estimate of the phase noise is obtained using pilot symbols inserted in the data sequence. The estimate is iteratively improved by exploiting also soft decisions from the data symbols, obtained in a previous iteration. We demonstrate that the resulting (linearized) mean-square estimation error consists of two contributions: a contribution from the additive noise, that equals the Cramer-Rao lower bound, and a noise-independent contribution that results from the phase noise modeling error. Performance can be optimized by a proper selection of the number of DCT coefficients in the expansion of the phase noise. Iterative soft-decision-directed phase noise estimation yields a considerable performance improvement as compared to estimation using only pilot symbols.