An Efficient CS-Based and Statistically Robust Beam Alignment Scheme for mmWave Systems

Millimeter-Wave (mmWave) communication has come into the spotlight as an enabling approach for next generation wireless networks. Communication at mmWave is however challenging due to the large path loss and limited power. This implies that antenna arrays with large directional gain are required both at the Base Station (BS) and the user sides. Finding the strongest narrow beam pair connecting the BS and the user is referred to as Beam Alignment (BA). In this paper, we propose an efficient BA scheme for multi-user systems via estimating the second order statistics of the channel. In the proposed scheme, the BS probes the channel in the downlink letting each user estimate its own channel, where all the users within the BS coverage are trained simultaneously. We formulate the channel estimation at the user side as a Compressed Sensing (CS) of a non-negative sparse vector and use the recently developed Non-Negative Least Squares (NNLS) technique to solve it efficiently. We evaluate our method via numerical simulations and compare it with other competitive algorithms. It has been verified that the proposed approach incurs less training overhead, exhibits higher efficiency in multi-user scenarios, and is highly robust to fast time-varying channels.