Enhanced sparse Bayesian learning-based channel estimation with optimal pilot design for massive MIMO-OFDM systems

The pilot contamination problem creates a limitation to the potential benefits of massive multiple input multiple output (MIMO) systems. To mitigate the pilot contamination, in this study, the authors propose a novel channel estimation for massive MIMO systems, using sparse Bayesian learning (SBL) based on a pattern-coupled hierarchical Gaussian framework. In the proposed technique, the sparsity of each channel coefficient is controlled by its own hyperparameter and the hyperparameters of its immediate neighbours. The simulation results show that the channel coefficients can be estimated more efficiently in contrast to the conventional channel estimators in terms of channel estimation with pilot contamination. Furthermore, they derive the mean square error (MSE) analytical expression for the proposed technique and based on that MSE expression, a pilot design criterion is proposed to design the optimal pilot to improve the estimation accuracy of the proposed algorithm using the Lagrange multiplier optimisation method. Results show that they can reduce the MSE of the SBL estimator by employing the optimal pilot sequence.