Spatial Channel Covariance Estimation for mmWave Hybrid MIMO Architecture


Sungwoo Park and Robert W. Heath, Jr.


to appear in Proc. of Asilomar Conf. on Signals, Systems, and Computers, Nov. 2016.


Instead of full channel state information, spatial channel covariance information can be used to design the analog part of mmWave hybrid MIMO systems. The problem is that the hybrid MIMO architecture makes it challenging to estimate the spatial channel covariance matrix because the estimator in the baseband can only see the low-dimensional projections of the original channel itself. In this paper, we propose two key ideas for developing the covariance estimation techniques based on compressive sensing techniques. One is to use the Hermitian property of the covariance, and the other is to use time-varying analog combining matrices to effectively extend the measurement size. We also propose an algorithm incorporating these two ideas.