Beamspace Joint Azimuth, Elevation, and Delay Estimation for Large-Scale MIMO-OFDM System

Due to the capability to separate the line-of-sight signal from multipath signals in both time-space domains, the joint azimuth, elevation angle and time delay estimation technique is of great importance in the Internet of Things. However, as mainstream devices develop toward the large-scale multiple-input and multiple-output (MIMO) system, real-time processing gradually becomes computationally impractical for traditional element-space three-dimensional (3-D) joint angle and delay estimation (JADE) methods. In this paper, based on the measured channel state information acquired from a large-scale uniform rectangular array-orthogonal frequency division multiplexing (OFDM) system, a computationally efficient 3-D beamspace JADE algorithm is proposed. Firstly, we develop a method to select the beam that contains the line-of-sight path as the optimal beam. Then, for the parameter estimation, we transform the channel state information into the beamspace by utilizing the discrete Fourier transform sequence, and propose a modified 3-D beamspace matrix pencil (MP) algorithm only with the optimal beam and its adjacent beams, which contributes to conspicuous computational savings. Moreover, the estimation of delay, elevation and azimuth for the line-of-sight path are paired automatically with only one eigenvalue decomposition, and the multi-dimensional grid search is avoided. Experiment results demonstrate that the proposed approach could correctly select the optimal beam with high probability, and its parameter estimation accuracy is superior to the state-of-the-art JADE techniques while significantly reducing the computational complexity.