Joint Channel Estimation and MAP Detection of Probabilistically Shaped QAM

This paper introduces a novel blind solution for joint channel estimation and data detection of convolutive mul-tichannel communications systems where Probabilistic Shaping (PS) is used. The convolutive channel estimation is based on the Expectation-Maximization (EM) algorithm, and data detection is achieved by re-utilizing the probabilities obtained within the EM framework, based on a maximum a posteriori (MAP) estimation. This work shows that, even if the source kurtosis is close to the Gaussian one, the blind estimation is still possible via the proposed method contrary to other existing HOS (Higher Order Statistics) based methods. Simulation results show that our algorithm provides an interesting channel estimation accuracy and a much better performance in terms of symbol error rate (SER) as compared to Hyperbolic-Givens multi-modulus algorithm (HG-MMA) which is based on multi-modulus criterion.