Parametric Phase Tracking via Expectation Propagation
share
In this work we propose a simple algorithm for signal detection in a single-carrier transmission corrupted by a strong phase noise. The proposed phase tracking algorithm is formulated within the framework of parametric message passing (MP) which reduces the complexity of the Bayesian inference by using distributions from a predefined family; here, of Tikhonov distributions. This stays in line with previous works mainly inspired by the well-known Colavolpe-Barbieri-Caire (CBC) algorithm which gained popularity due to its simplicity and possibility for decoder-aided operation. In our work we leverage the simplicity of the MP characteristic of the CBC algorithm and combine it with the principles of the expectation propagation (EP). In this way we notably improve the performance of the phase tracking before the decoder's feedback can be even considered. Further, in the spirit of the joint decoding and phase tracking, the EP algorithm can be integrated in the decoding loop; then, not only it outperform the reference scenario of the exact (discretized) message passing – a results that was not yet shown in the literature, but it requires much lower complexity than the state-of-the art algorithms.
READ FULL TEXT
