Sparse channel estimation based on compressive sensing kalman filter with limited feedback

Abstract

In this paper, the adaptive pilot signal sequence design for a large millimeter-wave (mmWave) multiple-input single-output (MISO) downlink channel estimation to minimize normalized mean square error (NMSE) is considered. By developing the sparsity in a large mmWave MISO downlink channel and the time dependency of a virtual channel vector (VCV) sequence, we reformulate designing the training beam sequence and channel estimation problem as the compressive sensing on kalman filter (KF-CS) that finds non-zero entries in a one-dimensional VCV. Under the KF-CS framework, feedback is used for designing optimal adaptive pilot signal sequence in an aspect of minimum mean square error (MMSE) with low computational complexity for a practical issue. To evaluate the performance of the proposed pilot signal design method, simulation results are provided and these shows the proposed method yields the better performance than the KF-CS in channel estimation.