Semi-blind MAP channel estimator for pilot decontamination

In this paper, a novel semi-blind channel estimator is introduced, with the ability of separating two or more independent wireless channels that are superimposed due to pilot contamination. Key novelties of this work include: a) exploiting the joint probability distribution of information and channel to facilitate the multi-channel separability of the maximum à posterior (MAP) channel estimator; b) exploiting the correlation between symbol detection results at a multi-antenna receiver to further improve the channel estimation accuracy. The mean-square error (MSE) of the proposed channel estimator is derived in a closed form, which exhibits the relationship between the pilot decontamination capacity and the modulation size as well as the data observation length. Moreover, simulation results demonstrate close MSE performances between the proposed estimator and the MAP estimator with orthogonal pilots.