DOA Clustering-based Channel Estimation for Mitigating Pilot Contamination in Massive MIMO System

Abstract

Massive multiple-input multiple-output (MIMO) has been considered to be one of the most significant technologies for next-generation communications, which is, however, challenged by the fundamental issue of pilot contamination. In this paper, the direction of arrivals (DOAs) difference caused by the user positions in different cells is utilized to separate the contaminated channels thus can improve the performance of channel estimation in the massive MIMO uplink transmissions. We first adopt a high-resolution subspace method to estimate the DOAs of resolving paths. Then, we use the K-means clustering method to distinguish the DOAs of the users in the target cell. Finally, we obtain the least squares (LS) estimation of the corresponding complex channel gain from polluted channels and estimated DOAs to estimate the transmit channel. Simulation results demonstrate that the proposed channel estimation scheme outperforms the traditional pilot-based method in terms of the mean-squared error (MSE) without requiring neither the statistical information about the channels nor any coordination between cells, but only needs the DOAs which can be obtained with low cost at the base station (BS).