Blind Channel Estimation for Millimeter Wave Uplink Systems with Unknown Number of Users

Millimeter wave (mmWave) communication is a promising technology in the 5G and beyond communications. Because of the ultra-high frequency, the mmWave communication suffers from significant path loss and the number of paths from users to the base station (BS) is much fewer than that in the conventional wireless communications systems. These problems can be mitigated by exploiting massive multi-input multi-output (MIMO) and beamforming techniques. To obtain the channel information for MIMO systems, the pilot-based channel estimation may incur a large overhead as the numbers of users and antennas increase. In this paper, we propose a blind channel estimation method for mmWave uplink that avoids using pilot signals. With the number of users unknown, we first estimate the number of users by the rank of the correlation matrix of the signal received at the base station (BS). Then, the compressed sensing (CS) technique is utilized to estimate the direction-of-arrival (DOA) of each path. At last, we take advantage of the special structure of the fading channel matrix to eliminate the spurious DOAs and estimate the fading gain of each path at the same time. A special case with only one path from each user to the BS is also discussed, and we exploit the multiple signal classification (MUSIC) algorithm to reduce the complexity. A non-uniform antenna array called the nested array is utilized to increase the array aperture for both the single-path per user and the general cases.