Downlink channel estimation based on block-structured compress sensing for multi-cell massive MIMO systems

The problem of channel estimation in multi-cell massive MIMO systems is addressed in this paper. The channels observed different antennas in the same cell have approximately sparsity and common support, and the sparsity and common support properties draw more and more attentions when it comes to the efficient estimation of large number of channels in massive MIMO systems. The short channel coherence times limits the upper length of the pilot sequence used for channel estimation. So it is impossible to allocate the unique pilot sequence for the different users in all the cells and the pilot sequence is reused in different cells which results in pilot contamination. To avoid the pilot contamination in multi-cell, we use the hadamard matrix to orthogonalize the pilot in adjacent cell. The coordinated approach improves channel estimation and eliminates the pilot contamination in adjacent cell. Extensive simulation results are provided to demonstrate the performance of the proposed algorithms.