Channel estimation by reduced dimension decomposition for millimeter wave massive MIMO system

Abstract

In this paper, a channel estimation method based on reduced dimension decomposition for millimeter-wave massive MIMO systems is proposed. For the sake of obtaining the high accuracy of the estimation, we decompose the channel matrix estimation into angle information and channel gain information estimation. The received signals are decomposed by dimensionality reduction so that the angles of the receiving end and transmitting end are separated. The sparse signal recover (SSR) scheme is used to acquire the initial sparse support set. Then, the off-grid error is regarded as the adjustment parameter. Using the orthogonal relationship between the signal subspace and the noise subspace, we gradually approximate the true discrete grid using Taylor’s formula. Finally, the path gain is estimated by the least squares estimation (LSE) algorithm. The benefit of the proposed method is the reduction of the training resources and costs. More importantly, to verify the estimated performance, we obtain the normalized mean square error (NMSE) performance of the channel matrix estimation and the achievable spectral efficiency (ASE) performance under the estimation scheme. Simulation results indicate that the proposed method has the effectiveness and superiority.