Channel Estimation with Pilot Contamination in Mutli-Cell Massive MIMO systems

Abstract

This paper compares the performance of three estimators for the Multi-Cell (M-C) Massive MIMO (m-MIMO) system with Pilot Contamination (PC) using the Mean Square Error (MSE) metric. The Least Square channel Estimator (LS-E), the current ideal Minimum Mean Square Error channel Estimator (MMSE-E), and the Maximum Likelihood channel Estimator (ML-E) are the compared estimators in our work. The LS-E undergoes substantial performance degradation, whereas the ideal MMSE-E gives much-enhanced performance over the LS-E. The ideal MMSE-E assumes perfect knowledge of cross cell Large Scale Fading Coefficients (LSFC), which is counting as an unjustifiable assumption in practical use. We decode the problem created by the ideal MMSE-E to adopt the ML-E by replacing the LSFC and the noise power in the ideal MMSE-E. The ML-E get over the issues of both discussed estimators and provides estimation performance close to that of the ideal MMSE-E. The MSE analytical expression for the ML-E is also calculating.