A Semi-Blind Channel Estimation Algorithm for Massive MIMO Systems

In this letter, we propose a semi-blind iterative space-alternating generalized expectation maximization (SAGE)based channel estimator for massive MIMO systems. The method updates the pilot-based minimum mean square error (MMSE) estimate iteratively with the help of the SAGE algorithm. Pilotbased estimators require additional pilot symbols for enhancing accuracy, whereas the proposed estimator uses data symbols. Several of the existing estimators assume complete information on large scale fading coefficients of the interfering cells which requires heavy overhead. However, the proposed estimator solves this problem by an estimate obtained from the received samples. Further, we show that the proposed estimator converges almost in one iteration, and achieves appreciable improvement over the existing pilot-based and data-aided estimators in a pilot contaminated scenario.