Pilot decontamination in wideband massive MIMO systems by exploiting channel sparsity

This paper strives to reduce pilot contamination, a bottleneck for massive multiple-input multiple-output (MIMO) systems, by exploiting channel sparsity. Considering that typical wideband massive MIMO channel is correlated in both space and frequency domains, we employ Karhunen-Loéve Transform (KLT) and Discrete Fourier Transform (DFT) to capture the hidden sparsity of the channel. KLT basis is optimal in extracting the uncorrelated information from channel, but requires channel statistical information. As a suboptimal alternative, DFT basis can be determined without channel statistics, which is more viable for practical use. By representing the channel with DFT basis, we find that the subspaces of the desired and interference channels are approximately orthogonal, even when the number of antennas is not so large. Inspired by this observation, we propose a pilot decontamination method, where a pilot assignment policy is designed to help identify the subspace of the desired channel, and a desired channel subspace aware least square channel estimator is derived to remove the pilot contamination. The proposed method does not need channel statistics and pilot co-ordination. By exploiting channel stationary, the method does not introduce extra training overhead. Simulation results demonstrate substantial sum rate gain of the proposed method over existing methods.