Sparse Channel Estimation Using Order Extension for Frequency and Time Selective MIMO Wireless Communication Networks

摘要

A channel matrix is estimated for a channel subject to frequency and time selective fading in a multiple-input multiple-output (MIMO) wireless communication network. The method uses a recursive least-squares (RLS) process and an expectation-maximization (EM) process, based on a L₁-L_q regularized sparse regression. The method significantly improves an accuracy of the channel estimation by using high-order matrix extension to obtain covariance matrices. The covariance matrices and an initial estimate are refined by a sparse EM to obtain the final estimate of the channel.