Channel Estimation Based Compressive Sensing in Millimeter Wave MIMO systems

摘要

Due to the sparse nature of the Millimeter-Wave (mmWave) multiple-input multiple-output(MIMO) channel, applying Compressive Sensing (CS) strategy to estimate the channel is advisable. However, when the transmit/receive angles (TRAs) in the channel are adjacent, the correlation between TRAs is too strong to seperate them from each other in the channel estimation process, which will severely degrades the channel estimation performance. Aiming at the special situation, firstly, we design two redundant dictionaries by finely quantizing the columns (angles) of DFT matrix, and manufacture a three-dimensional diagram, whose X, Y and Z axis correspond to the receiving angle, the transmitting angle and the correlation value between the vectors, namely TRAs, and the received joint signal. Secondly, we apply Orthogonal Matching Pursuit (OMP) to roughly estimate the TRA, then according to the positions and widths of the peaks, judge whether the actual TRA are neighboring. Finally, iteratively search the actual TRA in the smaller scope and their path's gains. Simulation results verify the better performance of the proposed scheme than its counterparts in the special situation.