Compressive Sensing for Indoor Millimeter Wave Massive MIMO : (Invited Paper)

Abstract

Under sparse channel assumptions, channel state information for the massive MIMO uplink can be effectively estimated without sampling every antenna. Assuming a slow flat fading multipath channel, orthogonal pilot signals, and a uniform rectangular array, channel estimation is performed by leveraging sparsity in the spatial and pilot code domains to reconstruct the channel to all antennas. Results of sampling from 25 percent of a 32 by 64 element massive MIMO array during the uplink piloting phase are presented. The sum rate achieved by using compressed sensing and sparse recovery channel estimates exceeds that achieved with the least squares channel estimate for the same number of sampled antennas.